Back to skill

Security audit

UAV Farm Health Index Map | 无人机农田健康指数图生成

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious with high confidence because scanners report behavior that goes beyond or differs from its stated crop imagery analysis, including account and identity handling, broad credential-bearing requests, and analysis that may run in an export-only mode.

Consider not installing until the account and credential behavior, remote destinations, and upload disclosures are clarified and the export-only mode is fixed. Skillspector completed most checks but marked coverage partial because a referenced local path was missing, so the evidence is incomplete.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/smyx_uav_farm_health_index_map_analysis.py
Finding

CLI identity initialization discards the upstream sender ID and reuses a shared account

Content
View full analysis

Impact

A requester can receive analysis records and report links created through the shared fallback account by other users. New analyses can also be attributed to that same account.

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (26)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied chunk is a user-database persistence component, not an agricultural image-processing implementation. It defines a sys_user model containing usernames, real names, email addresses, demographic fields, token, and open_token, and provides methods to read and modify these records. DAO initialization also creates database tables and adds missing user columns. No UAV imagery ingestion, vegetation-index computation, AI inference, or heatmap generation appears. Although database access can support a larger application, the user/account-management behavior shown here has no demonstrated connection to the declared crop-health workflow. This assessment is limited to the supplied code; it does not establish malicious intent or external transmission of stored data.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied file implements an account-aware API utility layer rather than the declared agricultural analysis. Generic file, JSON, date, and polling helpers could be supporting implementation details, and their presence alone would not establish a mismatch. However, silent account provisioning, identity-file access, credential-bearing requests, and billing handling are substantial behaviors not represented in the description. This assessment is limited to the supplied chunk: it does not establish that credentials are actually sent to an untrusted destination or that agricultural functionality is absent elsewhere.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The script explicitly presents itself as a video-analysis tool, documents MP4 inputs, and exposes a history-listing operation tied to an open ID. These capabilities materially differ from the declared agricultural image-processing workflow. The underlying skill implementation is not supplied, so its analysis internals cannot be established; nevertheless, the visible video-oriented interface and history retrieval demonstrate a mismatch with the declared purpose.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The supplied code does not implement the declared agricultural analysis workflow. The only substantive method is an inactive and broken generic agent-chat wrapper; it neither invokes an agent nor processes imagery. Behavior inherited from BaseSkill and ApiService cannot be determined from this chunk, so this assessment is limited to the visible implementation. No active subprocess execution or external data transmission is shown.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
86% confidence
Finding

The visible input contract explicitly refers to local videos and network video URLs and sends remote inputs as videoUrl. Its main operations are API submission, polling, response formatting, and report listing/export. These do not substantiate the specialized agricultural capability described. File reading and API calls alone are not treated as mismatches, since they could support the declared purpose. However, the visible video-oriented workflow materially differs from the stated orthophoto/mosaic analysis workflow. The imported API implementations are unavailable, so specialized agricultural processing in the backend cannot be ruled out.

Content

No source excerpt is available for this finding.

YARA rule 'agent_skill_mcp_tool_poisoning_metadata': MCP/tool metadata poisoning indicators in tool schemas or skill manifests [agent_skills]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · SKILL.md (reported line 3)May include surrounding context.

md
---
name: "smyx-uav-farm-health-index-map-analysis"
description: "Using multispectral or high-resolution RGB cameras mounted on agricultural UAVs to capture orthophotos or mosaics of farmland, AI models compute vegetation indices (e.g., NDVI, NDRE) and generate a farm health-index heatmap, where colors distinguish crop vigor (red = poor, yellow = medium, green = healthy). | 通过农业无人机平台搭载的多光谱或高分辨率RGB相机,采集农田的正射影像或拼接图,利用AI模型计算植被指数(如归一化植被指数NDVI、归一化红边指数NDRE等),生成农田健康指数热力图,用颜色区分作物长�

The request helper can send shared account credentials to arbitrary HTTP or HTTPS destinations.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

The helper accepts arbitrary absolute HTTP or HTTPS URLs and automatically attaches shared access tokens, an API key, and authorization credentials without checking the destination. A caller-influenced destination could therefore receive reusable credentials, and plain HTTP additionally exposes them in transit; crop analysis provides no justification for this unrestricted credential forwarding.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, file_read, file_write, network, shell.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Activation rules allow non-agricultural drone imagery and generic image-analysis requests to trigger farm-health analysis.

Medium
Category
Trigger Abuse
Confidence
93% confidence
Finding

The default trigger includes requests to analyze UAV orthophotos, mosaics, or videos without explicitly requiring agricultural content or farm-health intent. This can route unrelated drone imagery into the cloud-analysis workflow, exposing sensitive imagery and producing inappropriate agricultural assessments.

Content

No source excerpt is available for this finding.

The --export-env-only option promises not to execute analysis, but normal analysis still runs.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The export-only flag is parsed but never checked, so supplying it with an input still initializes identity and executes analysis. A caller relying on the explicit promise of no analysis can unintentionally submit imagery to the service or incur service usage.

Content

No source excerpt is available for this finding.

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · skills/smyx_analysis/scripts/skill.py (reported line 28)May include surrounding context.

python
result_json = JsonUtil.parse(result_json_pure_text, result_json_pure_text)

        result_json_common_ai_response = result_json.get("commonAiResponse") if isinstance(result_json,
                                                                                           dict) else result_json
        if result_json_common_ai_response:
            result_json = result_json_common_ai_response

Whitespace Padding

Medium
Category
Prompt Injection
Confidence
70% confidence
Finding

Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.

Content

Scanner excerpt · skills/smyx_analysis/scripts/skill.py (reported line 33)May include surrounding context.

python
result_json = result_json_common_ai_response

        result_json_health_ai_response = result_json.get("healthAiResponse") if isinstance(result_json,
                                                                                           dict) else result_json
        if result_json_health_ai_response:
            result_json = result_json_health_ai_response

Local video contents are submitted to the analysis API without disclosure of the upload.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

Local file contents are read in full and passed to the remote analysis operation without an upload disclosure or consent check at this boundary. Although the broader skill mentions cloud APIs, the local-file workflow does not clearly explain that potentially sensitive farm imagery and embedded metadata leave the machine.

Content

No source excerpt is available for this finding.

The documented export-only mode still executes analysis.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The documented export-only option has no corresponding control-flow branch, so valid input still reaches the analysis call. This violates a caller's expectation of a non-executing mode and can cause unintended remote data transfer or service usage.

Content

No source excerpt is available for this finding.

The skill includes persistent user and credential management unrelated to its declared crop-health analysis purpose.

Medium
Category
Excessive Agency
Confidence
93% confidence
Finding

L448–L465 define a user table containing personal information and token/open_token fields, and L470–L473 expose it through UserDao's inherited read, create, update, and delete operations. DAO initialization also creates and migrates this user table (L184–L215), introducing identity-store modifications that are not justified by the manifest's UAV imagery, vegetation-index computation, or heatmap-generation scope.

Content

No source excerpt is available for this finding.

ai_chat documents an agent invocation but never invokes an agent or returns analysis.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The docstring at L26-L34 states that this method invokes openclaw agent through subprocess and returns an analysis result or session ID. However, command construction and execution are commented out, and the empty dictionary at L68 causes result.stderr at L70 to raise an exception that is caught and suppressed, so the method always returns None rather than performing the documented analysis.

Content

No source excerpt is available for this finding.

FileUtil.open silently truncates existing files without disclosing its destructive behavior.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

Opening a path in 'w' mode immediately erases any existing contents. This generically named helper has no confirmation, action log, docstring, or comment explaining that it overwrites files; exception tracing does not disclose successful truncation.

Content

No source excerpt is available for this finding.

Ordinary API requests can silently register an external account and persist its identity and tokens.

Medium
Category
Excessive Agency
Confidence
95% confidence
Finding

An ordinary request can invoke phoneLogin with silent registration enabled, transmit the current identity, and persist returned account information and tokens. This creates externally visible and durable account state without an explicit enrollment step, which is an unexpected privacy and credential-management side effect for a crop-health skill.

Content

No source excerpt is available for this finding.

HTTP requests can silently trigger remote account registration and transmit the current identity.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The generic HTTP helper conceals a separate account-registration request using silent=1 and register=1, sending the current identity as both openId and mobile. This is a real undisclosed side effect rather than a requirement inherent to vegetation-index computation, and overlaps the account-enrollment finding.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
93% confidence
Finding

This POST transmits the selected identity and skill-hub source to an external login/registration endpoint as part of implicit account provisioning. External API traffic is not inherently unsafe, but this transmission is security-relevant because it occurs automatically without a visible consent boundary; the source does not establish a malicious destination.

Content

Scanner excerpt · skills/smyx_common/scripts/util.py (reported line 561)May include surrounding context.

python
"source": ConstantEnum.DEFAULT__SKILL_HUB_NAME
            }
            try:
                _response = requests.post(_url, json=_data)
                if _response.status_code == 200:
                    _response_json = _response.json()
                    if _response_json and _response_json.get("success"):

Configuration loading silently creates directories and files, including during module import.

Low
Category
Output Handling
Confidence
96% confidence
Finding

YamlUtil.load creates parent directories and writes a YAML file when the requested configuration is absent, without a log, prompt, docstring, or explanatory comment disclosing this side effect. BaseEnum.init_subclass invokes this loader when subclasses are defined (L052–L059), so importing the module can modify the filesystem unexpectedly.

Content

No source excerpt is available for this finding.

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Confidence
50% confidence
Finding

Dynamic getattr() with a non-literal attribute name can access arbitrary object attributes, potentially bypassing access controls.

Content

Scanner excerpt · skills/smyx_common/scripts/dao.py (reported line 243)May include surrounding context.

python
username = getattr(model, "username", None)
            if username:
                column_names = self.__model__.__table__.columns.keys()
                update_data = {key: getattr(model, key) for key in column_names if key != "username"}
                return self.update_by_username(username, **update_data)

            return None

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Confidence
50% confidence
Finding

Dynamic getattr() with a non-literal attribute name can access arbitrary object attributes, potentially bypassing access controls.

Content

Scanner excerpt · skills/smyx_common/scripts/dao.py (reported line 353)May include surrounding context.

python
column_names = self.__model__.__table__.columns.keys()

            for key in column_names:
                value = getattr(model, key)
                setattr(instance, key, value)

            session.commit()

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Confidence
50% confidence
Finding

Dynamic getattr() with a non-literal attribute name can access arbitrary object attributes, potentially bypassing access controls.

Content

Scanner excerpt · skills/smyx_common/scripts/dao.py (reported line 326)May include surrounding context.

python
if filters:
                for key, value in filters.items():
                    query = query.filter(getattr(self.__model__, key) == value)

            if offset:
                query = query.offset(offset)

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Confidence
50% confidence
Finding

Dynamic getattr() with a non-literal attribute name can access arbitrary object attributes, potentially bypassing access controls.

Content

Scanner excerpt · skills/smyx_common/scripts/dao.py (reported line 441)May include surrounding context.

python
if filters:
                for key, value in filters.items():
                    query = query.filter(getattr(self.__model__, key) == value)

            return query.scalar()
        finally: